Google Cloud’s Distributed Relational Database Spanner Goes Global
Google Cloud Spanner, the Google Platform’s distributed relational data store service released earlier this year as a regional offering, has gone global.
The original release offered data synchronization across three availability zones. With Google’s release of multiregion configurations, it offers synchronization across multiple availability zones and across regions and continents.
“For customers that are global and have a global user base and want a global database, to make sure their mission-critical data is available, up-to-the-minute accurate, that’s the option that Cloud Spanner provides,” said Deepti Srivastava, Google product manager for Cloud Spanner.
Built from the ground up, Spanner always had this global capability, Srivastava said, but the company wanted to get customer feedback on it.
“We wanted to make sure we weren’t just releasing cool tech for tech’s sake, but that we were actually solving customer problems, that there was an actual need for customers to do this,” she said.
In addition, it’s offering a 99.999 percent availability service-level agreement with no planned downtime — less than five minutes of downtime a year. That indicates the company’s level of confidence in the system’s reliability, she said.
The company has touted the database as offering the familiarity of relational database semantics with non-relational scale and performance. One of the most talked-about features has been strong consistency for transactions provided through GPS receivers and atomic clocks, timekeeping technology called TrueTime.
In effect, Spanner creates one global relational database, not just a feature that copies data around the world. But just because it offers that capability doesn’t necessarily mean every company should use it.
The varied data governance regulations around the world, including the General Data Protection Regulation (GDPR) coming to the European Union in May 2018 — which experts say most companies are woefully unprepared for — pose huge challenges for use of such a database.
It’s one of the reasons Spanner was released early on as merely a regional offering, Srivastava said.
“If you have data governance regulations, we always recommend you stay with a regional product because then you’re guaranteed to have data replicated only within that region,” she said.
Google built Spanner in-house to power services such as AdWords and Google Play after the manual sharding required in its MySQL implementations became too unwieldy. It released Cloud Spanner after almost 10 years of internal use.
In the coming months, note-taking site Evernote plans to replace more than 750 MySQL instances with single Cloud Spanner instance to manage more than 8 billion plus pieces of its customers’ notes.
For developers, Spanner offers a Java Database Connectivity (JDBC) driver and software development kits in languages such as Java, Go, Python and Node.js. A Microsoft Open Database Connectivity (ODBC) driver and C# SDK are in the works.
CockroachDB, recently announced YugaByte and the Chinese project TiDB also aim to create SQL databases as scalable as NoSQL systems while maintaining ACID transactions, but they have not duplicated TrueTime.
Google Cloud is a sponsor of The New Stack.
Feature image via Pixabay.